(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: studies have reported that a phase-space-based source model based on a complete Monte Carlo simulation of an X-ray tube is sufficiently accurate for calculating 3D dose distributions. However, the large amount of phase-space data is unsuitable for a graphics processing unit (GPU) based parallel computing because of the data transfer overhead between a CPU and GPU memory. Hence, we established an analytic source model for GPU-based dose calculation.
Methods: kilovoltage tube geometry was simulated using EGSnrc/BEAMnrc code. The reference phase-space file stores the particle positions, directions, and energy. Particles in the file were classified into two groups: primary and secondary photons. The source model parameterized the probability densities of each parameter, such as the energy spectrum, fluence distribution, and direction cosine distribution. To validate the accuracy of our source model, we used it to calculate the dose distributions of each imaging mode and compared the results with those directly calculated using a reference phase-space file.
Results: primary photon spectra varied with the off-axis distance, whereas the secondary photon spectra were consistent in the field (standard deviations of mean photon energies: primary, 4.1 keV; secondary, 1.7 keV). This suggests that the secondary photon is a single energy spectrum. The percentage of secondary photons in total fluence increased in bowtie filter fields (open field, 4.7%; full bowtie field, 15.9%; half bowtie field, 15.2%). Secondary photons were important components in the bowtie-filtered fields. The dose distributions between the two source models demonstrated good agreement. The average dose difference was within 1.0% for the PDD beyond the depth of the maximum dose and the OAR in the inner and outer beam regions.
Conclusion: results demonstrated the efficacy of our source model in terms of accurately representing a reference phase-space file. Thus, it can achieve high efficiencies in GPU-based dose calculation engines.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
